RESEARCH Open Access A multi-layered OFDM system with ......RESEARCH Open Access A multi-layered OFDM system with parallel transmission for multicell cooperative cellular networks

RESEARCH Open Access

A multi-layered OFDM system with paralleltransmission for multicell cooperative cellularnetworksJiaxin Yang1, Xianbin Wang1*, Sung Ik Park2 and Heung Mook Kim2

Abstract

Recent development in multicell cooperation poses significant technical challenges to the design of robust andflexible transmission techniques. A new multi-layered orthogonal frequency-division multiplexing (ML-OFDM)system is proposed in this article to provide a dynamic platform for multicell cooperation with efficient basestation coordination capability. The proposed enhanced layers (ELs), which are overlaid with the cellularcommunication data (the base layer) in both frequency and time domains, can be used for several specificpurposes indispensable to multicell cooperation. It provides an efficient way of sharing the necessary information,e.g., channel state information, user data and other transmission parameters, between the collaborative BSs withoutthe requirement of additional signaling or control channels. Overall network efficiency is substantially enhanceddue to the reduction of radio resource overhead. Furthermore, cross BS synchronization and multimedia broadcastmulticast service for next generation cellular networks can be simultaneously achieved by the proposed parallelorthogonal ELs. The transceiver design for the ML-OFDM system, particularly the modulation/demodulation of theELs and EL-induced interference cancelation is presented. Overall system performance is further optimized byproposing a power distribution scheme with a set of practical constraints. The performance of the ML-OFDMsystem is analyzed and verified through numerical simulations.

I. IntroductionThe ever-increasing demand for broadband mobile mul-timedia applications brings significant challenges to thedesign of future-generation cellular networks. Due tothe significant users’ requirements for mixed services,the concept of cellular hybrid [1-4] is becoming an intri-guing requirement to simultaneously support differentservices, e.g., the two main types: point-to-point unicast(conventional cellular communications) and point-to-multipoint broadcast. However, today’s design metho-dology of cellular networks is characterized by uncoordi-nated approaches in several aspects. On one hand, theintegration of unicast and broadcast services are usuallyrealized by utilizing separate wireless infrastructures ororthogonal multiplexing techniques including time/

frequency-division multiplexing (TDM/FDM),a leadingto inefficient hardware or resource utilization.On the other hand, the conventional cellular commu-

nications based on single cell processing (SCP), havevery limited sharing of spectrum resources due to theresultant large inter-cell interference, and therefore, pre-venting the potential enhancement of network through-put and coverage [5]. Although the SCP generally servedwell in the past 2G/3G networks, the growing popularityof high-speed wireless applications in recent years posesa looming challenge due to the performance limitationof the existing methodology, necessitating a new trans-mission paradigm referred to as multicell cooperation,which exploits the inter-cell interference cooperativelyby enabling joint signal processing among several inter-fering base stations (BSs).Multicell cooperation, sometimes also known as dis-

tributed antenna system or multicell multiple-input-mul-tiple-output (MIMO), is a revolutionary technique whichaims to eliminate the capacity-limiting factor of conven-tional cellular networks and remarkably improve the

* Correspondence: [email protected] Centre for Information Engineering, Department of Electrical andComputer Engineering, The University of Western Ontario, London, Ontario,N6A 5B9, CanadaFull list of author information is available at the end of the article

Yang et al. EURASIP Journal on Wireless Communications and Networking 2012, 2012:103http://jwcn.eurasipjournals.com/content/2012/1/103

© 2012 Yang et al; licensee Springer. This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium,provided the original work is properly cited.

mailto:[email protected]

http://creativecommons.org/licenses/by/2.0

overall system performance [6-9]. This intelligent wire-less system prescribes coordinated signaling strategiessuch as power allocation, beamforming directions, userscheduling, and joint encoding/decoding of the trans-mitted/received signals at the BSs depending on therequired levels of cooperation [6]. Recently, it hasattracted lots of attention from both industrial and aca-demic communities. For instance, the 3GPP LTE-Advanced [10] standard, where the network coordina-tion is known as coordinated multi-point (CoMP) trans-mission has been calling for standardization of signalingschemes for this technique since September of 2010 forpossible consideration in future Releases of LTE-Advanced. Although many pioneering studies have beendone in the literature, which evaluate the performanceof multicell cooperation through various information-theoretic models with simplified assumptions [11-21],the real implementation-related issues will still result insignificant technical challenges in the design of trans-mission schemes for this new technique.(1) Backhaul issues: Current multicell cooperation

techniques are enabled by the presence of a backhaulnetwork with unlimited capacity and free delay, whichconnects the BSs with each other or with a central pro-cessor. Since this assumption is quite unrealistic, spe-cially for large scale networks, the limited-capacitybackhaul within the pre-existing infrastructure may beunable to reliably transmit channel state information(CSI) and user data information among the collaborativeBSs. As a result, the desired transmission techniqueshould be capable of providing a robust signaling oftransmitting the required information between colla-borative BSs. The desired signaling can be used supple-mentarily to reduce the burden of the existing limitedbackhaul network or as stand-alone scheme when thebackhaul network is unavailable.(2) Network latency: Due to the large overhead of

global CSI/user data and the limited transmission cap-ability of the backhaul network, the distribution of thenecessary information among BSs must be achieved bywell-designed cross-layer algorithm including mediaaccess control (MAC) layer scheduling as well as physi-cal (PHY) layer transmission strategies [22-25]. Thecommunication between the PHY and upper layers pro-tocols, and the traffic routing will naturally bring exces-sive time delay, causing dramatic performancedegradation especially when the delay exceeds thecoherence time of the downlink channels.(3) BS synchronization: To guarantee the mitigation

of inter-cell interference, the desired signal componentstransmitted from different collaborative BSs to the targetmobile user must arrive synchronously. Efficient andaccurate cross BS synchronization is another

fundamental enabling technology for multicell coopera-tion since the imperfect timing advance will inevitablyhave negative impacts, e.g., power degradation of thedesired signal and additional intersymbol interference(ISI) [6]. Hence, tight synchronization between BSs byexploiting alternative signaling scheme is quite challen-ging, specially in situations where the global positioningsystem (GPS) signal is unreliable (e.g., indoor or denseurban areas).The motivation of this article is to address the afore-

mentioned challenges with the proposed multi-layeredorthogonal frequency-division multiplexing (ML-OFDM) system. OFDM is envisioned as a key technol-ogy for broadband wireless communications due to itshigh spectral efficiency, robustness to multipath distor-tions, and simple receiver design [26]; most of thebroadband systems (DVB-T, WiMAX and LTE) arealready OFDM-based, and therefore, we propose a ML-OFDM system which provides a robust, efficient andflexible platform specially tailored for the newly concep-tual multicell cooperative cellular networks. The princi-ple of the ML-OFDM system is shown in Figure 1,where the base layer (BL) provides conventional ortho-gonal frequency-division multiple access (OFDMA)-based unicast services for cellular users and theenhanced layers (ELs) offer several other importantfunctionalities for the cellular network. First, the pro-posed EL can provide a dedicated over-the-air linkamong different BSs of exchanging the available infor-mation including the CSI pertaining to all relevantdirect and interfering links, data symbols sent to thetarget mobile user and other transmission parameterssuch as power, beamforming coefficients, time slot, sub-carrier usage and etc. These information can be sentconcurrently with the user data-carrying signal (signalof BL) for dynamic BS coordination purposes. Suchcoordination protocol can be realized by solely exploit-ing the proposed EL or using the signaling to enhancethe pre-existing finite-capacity backhaul. Second,another highly desired feature called multimedia broad-cast multicast service (MBMS) for the upcoming 3GPPcan be supported by other parallel ELs. Target applica-tions include mobile TV, radio broadcasting, as well aspublic messaging and emergency alerts for all the silentreceivers within the cell. As the ELs and the BL areoverlaid across both the entire frequency and timedomains, the tedious procedure to establish separateinfrastructures or design orthogonal multiplexing couldbe avoided, which significantly reduces the implementa-tion cost and radio resource overhead. In addition,cross BS synchronization can be efficiently designedbased on the alternative EL. Each BS sends its uniquebeacon signal which carries timing and frequency


Page 2 of 18

information to its surrounding BSs. By detecting the sig-nals transmitted from other BSs and comparing thetiming and frequency information with its local refer-ence, the offsets are frequently compensated [27]. Themajor difference between the proposed ELs and conven-tional control channels for the aforementioned purposesis that the ELs can support multiple functionalitiessimultaneously without occupying additional radioresources.The rest of the article is structured as follows. The

principle and architecture of the proposed ML-OFDMsystem are presented in Sections 2 and 3. A new modu-lation scheme for the proposed ELs and the correspond-ing transmitter/receiver design are studied. Based on theinterference analysis for the proposed ML-OFDM, anefficient EL induced interference cancelation algorithmis also proposed. In Section 4, we analyze the systemperformance including the error probability of the pro-posed EL and its impact on the capacity of the BL.Based on these analysis, a power distribution scheme isproposed which optimizes the overall system perfor-mance with some practical constraints in Section 5.Simulation results are provided and discussed to evalu-ate and validate the performance and feasibility of theproposed system in Section 6. The article is finally con-cluded in Section 7.

II. Transmitter design for the proposed systemA. Overall signal structureThe transmitter’s block diagram of the proposed ML-OFDM system is shown in Figure 2a. Let X(k) denotethe kth user’s complex-valued data on the kth subcarrier

of the BL and N denote the total number of subcarriers.The corresponding OFDM block of the BL is given by

X =[X(0),X(1), ...,X(N − 1)

]. (1)

Without loss of generality, we assume a total of K ELsand the BL are overlaid across both the frequency andtime domains. The K ELs are designed to provide multi-ple functionalities including BS cooperation signaling,synchronization and MBMS, etc. The data streams onthese ELs are first modulated by the proposed scheme(described in the following subsection) and then super-imposed onto X with different power. Therefore, theoverall frequency-domain signal can be formulatedaccording to

S = X +K∑i=1

√PiEi, (2)

where Ei denotes the signal vector of the ith EL and Pidenotes its corresponding transmit power. Each time-domain data block is then generated by N point inversediscrete Fourier transform (IDFT),

s(n) =1√N

N−1∑k=0

S(k)ej2πknN , n = 0, 1, 2, ...,N − 1. (3)

Note that by assuming that the cyclic prefix (CP) inthe system is longer than the maximum channel delayspread, the transmitted symbols are free of ISI andtherefore, the insertion and removal of the CP will notbe included in the following discussions throughout thearticle.

Figure 1 Illustration of a cellular network with multicell cooperation and MBMS using the proposed ML-OFDM. (a) Snapshot of thecellular network, (b) The signal structure of the proposed ML-OFDM.


Page 3 of 18

B. Modulation of the ELsIn this subsection, we propose a modified code shiftkeying (MCSK) for the ELs. For the conventional CSK,M different cyclic phase shifts of a signature sequencewith length longer than M is employed as M-ary signal-ing to transmit data symbols [28-30]. The CSK can offervery high noise and interference immunity such thatextremely robust transmission can be achieved which ishighly desired by multicell cooperation and MBMSapplications. However, the main challenge of the con-ventional CSK is its limited data rate by the length ofthe signature sequence. Hence, we propose to increasethe data rate by using a shorter signature sequence such

that multiple data symbols can be transmitted withinone data block.Without loss of generality, we describe the modulation

of the jth EL. Denoting the signature sequence used onthis layer as z(j) with length M and assuming N/M = �

is an integer, we can collect total Ψ data symbols withinone data block with each symbol mapped to a modu-lated sequence. As shown in Figure 3, the input datastream is first grouped into k-bit

(k = log2M

)data sym-

bols and thus can be represented as

D(j) =[d(j)0 ,d(j)

1 , ...,d(j)�−1

], (4)

BinaryInformation

Source

M-QAMSymbolMapping

S/P

N pointIFFT P/S

MulticellCooperation/BroadcastInformation

Pseudo RandomSequenceGenerator

CSKModulationS/P

0

1

N-2

N-1

......

Multiplexing

1

K

...

1

K

...

D/A

Code 1 Code K... ...

0

1

N-2

N-1

......

S/P

TransmissionPower

Distribution

Sync&

ChannelEstimation

N pointFFT FEQ

QAMDemod

InterferenceCancellation

(a)

(b)

BSCooperation

BS

MS

Joint SignalProcessing

Output

Output

CSKDemod......

CSKDemod

CSKDemod

CSKDemod

1

2

K-1

K

Figure 2 Block diagram of the proposed ML-OFDM system. (a) Transmitter, (b) Receiver.


Page 4 of 18

with each data symbol dm denoted by

d(j)m =

[d(j)m,0, d

(j)m,1, ..., d

(j)m,k−1

]. (5)

According to the symbol value of d(j)m , the signature

sequence z(j) is cyclicly shifted by a unique phase Om

determined by the symbol value and denoted by z(j)Om.

The phase Om can take value of {0, 1, ...,M − 1} andtherefore, log2M bits can be transmitted within eachdata symbol and a total number of � · log2M withinone data block. The related transmission parameters ofthe MCSK are summarized in Table 1.After one-to-one mapping between each data symbol

and a modulated sequence, the output signal on the ithEL can be represented by

Ej =[z(j)O0

, z(j)O1, ..., z(j)O�−1

]. (6)

The above steps can be repeated for the other ELsexcept that the signature sequence z(j) is now replacedby z(i), i = 1,..., K, i = j for modulation purpose.

We now consider the design of the signaturesequences. Denote the set of signature sequences used bythe K ELs by,

Z ={z(0), z(1), z(2), ..., z(K−1)

}. (7)

Due to the superposition of parallel ELs, the mutuallayer interference is inevitably introduced. To eliminatethe impact, the signals on different ELs should bedesigned to be orthogonal. Therefore, the ideal design ofthe signature sequences should meet the following cri-teria,

M−1∑k=0

z(i)Om·(z(j)On

)∗={M, if i = j

⋂Om = On

0, if i �= j⋃

Om �= On. (8)

Eq. (8) indicates that ideally the signature sequenceshould be orthogonal to its cyclically shifted versionsand other sequences regardless of any shift in the setsuch that the mutual interference between different ELswill completely be eliminated. However, sequences withsuch perfect correlation properties do not exist inmathematics. Recently, Zadoff-Chu sequence hasreceived considerable attention and has been adopted by3GPP LTE air interface as primary synchronization sig-nal (PSS) and random access preamble (PRACH). Oneimportant property of Zadoff-Chu sequence is that ithas perfect cyclic auto-correlation and small cycliccross-correlation values. By assigning nearly-orthogonalsequences to different ELs, the mutual layer interferenceis significantly reduced while different ELs can be

11100101 11001001di

k bits per symboldj

k bits per symbol

Symbolvalue Oi

Symbolvalue Oj

Signature sequence

Zicorresponding to code phase

OiZj

corresponding to code phase Oj

Input data stream

1 2, , , ,i jZ Z Z Z ZZ Z Z, ,i j,,, ,,,Z Z Z, ,i j,,MultiplexingOutput signal

Figure 3 Illustration of the proposed MCSK modulation used by the ELs.

Table 1 Parameters of the MCSK

Parameters Values

Length of signature sequence MNumber of bits/Data symbol k = log2M

Number of sequences/OFDM block � = N/MTotal number of bits Ψ · k


Page 5 of 18

uniquely identified at the receiver. In addition, itimproves over other sequences due to its constantamplitude which reduces the complexity in the designof radio power amplifier. Therefore, it is preferred forCSK modulation and used as the signature sequence inthe proposed system.For clarity of exposition, an example is given where

N = 1024 and M = 64, then it is possible transmit 6bits for each sequence and a maximum number of 6 ×16 = 96 bits can be transmitted for the EL within onedata block. In the case of 3GPP-LTE evolved universalterrestrial radio access (EUTRA) air interface where thesymbol duration is 66.7 μs, data rate for this EL can beas high as Rb = 96 bits/66.7μs ≈ 1.4 Mbps. This rate issufficiently high for MBMS, i.e., video conferencingquality stream (128 - 384 kbps) and VCD quality stream(1.15 Mbps max), as well as information sharing formulticell cooperation purposes.

C. Advantages of the proposed modulationThe proposed MCSK is capable of providing a robust,efficient and flexible way to transmit additional data onthe ELs. Its advantages can be summarized as follows:

1) The MCSK offers significantly higher robustnessthan the conventional quadrature amplitude modula-tion (QAM) and based on which, effective EL-induced interference cancelation can be designed toreduce the impact of the ELs on the cellularcommunications.2) No dedicated radio resources are needed addition-ally as the transmitted signals on the ELs and thecellular users’ data are fully overlapped.3) The data rate of the ELs can be flexibly adjustedaccording to the system specified requirements. Forexample, in emergency communications whichrequires low data rate but very high reliability, eachdata symbol can be repeated by several times in onedata block, e.g., (4) can be reformulated as

D(j) =

⎡⎢⎣d(j)

0 , ...,d(j)0︸︷︷︸

R

, ...,d(j)�/R−1, ...,d

(j)�/R−1︸︷︷︸

R

⎤⎥⎦ . (9)

The benefit of this strategy is that at the receiver side,coherent combining can be performed prior to datadetection to enhance the signal-to-noise ratio (SNR) ofthe link. The same mechanism can also be applied tothe muticell cooperation, when the overhead of theshared CSI/user data is reduced due to the use of quan-tization techniques [31,32].

III. Receiver design for the proposed systemIn order to implement the proposed ML-OFDM system,some modifications are necessary to traditional OFDMreceiver. As the overlay ELs’ signals appear to be largeinterference to the BL’s signal, to guarantee the servicequality of the cellular communications, the first step is todemodulate the data on the ELs and the receiver is thencapable of removing the interference induced by the pre-vious demodulated ELs based on the estimated channeland the regenerated signals on the ELs.

A. Data detection of the ELsConsider a block fading multipath channel h = [h0, h1,h2, ..., hL-1] and its frequency response can be denotedby H whose elements are given by,

H(k) =L−1∑l=0

hlej2π lkN , k = 0, 1, ...,N − 1. (10)

The received signal after time to frequency domainconversion can be represented as

Y(k) = X(k)H(k) +k∑i=1

√PiEi(k)H(k) +W(k), k = 0, 1, ...,N − 1, (11)

where W(k) denotes the kth subcarrier additive whiteGaussian noise (AWGN) sample with zero mean andvariance σ 2

n . For analytical simplicity, we assume thatthe channel is estimated and compensated and there-fore, the signal after frequency-domain equalization canbe written as

S′(k) = X(k) +K∑i=1

√PiEi(k) +W ′(k), k = 0, 1, ...,N − 1,(12)

where W’(k) ≜ W(k)/H(k) represents the kth subcarrierAWGN scaled by the channel frequency response. To beconsistent with Section 2.B, we now discuss the demodula-tion of the mth data symbol of the jth EL. The cyclic phaseembedded in the sequence is detected by computing thefrequency domain cross-correlations between the corre-sponding signal segment and the local generated signaturesequence with all the possible cyclic phase shifts:

R(ϕ) =M−1∑k=0

S′ (k +mM)(z(j)Oϕ

(k))∗, Oϕ = 0, 1, ...,M − 1, (13)

where z(j)Oϕ,Oϕ = 0, 1, ...,M − 1 is the signature

sequence of the jth EL with all the possible cyclic phaseshifts.Substituting (12) into (13), (13) can be further

expanded as


Page 6 of 18

R(ϕ) =M−1∑k=0

[X(k +mM) +

K∑i=1

√Piz

(j)Om

(k) +W ′(k +mM)

](z(j)Oϕ

(k))∗

=M−1∑k=0

√Pjz

(j)Om

(k)(z(j)Oϕ

(k))∗

+K∑

i=1,i�=j

M−1∑k=0

√Piz

(j)Om

(k)(z(j)Oϕ

(k))∗

︸︷︷︸Mutual Interference between the ELs

+M−1∑k=0

[X(k +mM) +W ′(k +mM)

] (z(j)Oϕ

(k))∗

︸︷︷︸Base Layer’s Interference and Noise

=

⎧⎪⎪⎪⎨⎪⎪⎪⎩√Pj

M−1∑k=0

∣∣∣z(j)Om(k)∣∣∣2 + (m+1)M−1∑

k=mM

[X(k) +W ′(k)

] (z(j)Oϕ

(k))∗

if Om = Oϕ

mM−1∑k=(m−1)M

[X(k) +W ′(k)

] (z(j)Oϕ

(k))∗

if Om �= Oϕ

(14)

The mutual layer interference between the ELs can beassumed to be negligible due to the use of the ZadoffChu sequences when compared with the strong interfer-ence from the BL and the AWGN, which is

K∑i=1,i�=j

M−1∑k=0

√Piz

(j)Om

(k)(z(j)Oϕ

(k))∗ ≈ 0. (15)

The local signature sequence with cyclic phase shiftthat leads to the maximum correlation output is thephase encoded from the transmitted data bits,

O′m = argmax

Oϕ

{R(ϕ),Oϕ = 0, 1, ...,M − 1

}. (16)

With the one to one mapping between the decimalvalue of the input data symbol and the cyclic phase shiftO′

m, the original data d(j)m in (4) can be retrieved.

B. Interference cancelation of ELsThe superposition of ELs’ signals may cause large inter-ference to the demodulation of the OFDM data of theBL. However, due to the significantly large gap betweenthe operating SNR ranges of the ELs and the BL, the biterror rates (BERs) of the ELs can already be very low atthe SNR of traditional OFDM receiver such that theprobability that the regenerated ELs’ signals E’i = Ei isclose to unity. Hence, E’i can directly be subtractedfrom the received signal according to,

S′′ = S′ −K∑i=1

√PiE’i. (17)

In practical wireless communications, the interferencecancelation can be imperfect due to the channel estima-tion and data detection errors, resulting in certain resi-dual interference to the BL after the subtraction. If wedenote the estimated channel frequency response as H’and the residual interference can be written as

I =K∑i=1

(Ei · H’ − E’i · H’

)

=K∑i=1

(Ei · �H + �Ei · H − �Ei · �H),

(18)

where (·) denotes the element-by-element multiplica-tion. ΔH ≜ H - H’ and �Ei � Ei − E’i denote the errorof channel estimation and regenerated signal, respec-tively. Note that the last term in (18) is small in magni-tude and can be neglected. Then the variance of I canbe calculated as,

σ 2I ≈

K∑i=1

(Piσ 2

�H + 2Pe,iPiσ 2H

), (19)

where σ 2�H and σ 2

H denotes the variance of channel esti-mation error and the total channel power, respectively.Pe,i represents the symbol error rate (SER) of the ith EL.The residual interference is influenced by both channelestimation and data detection results. Nevertheless, as Pe,i is already sufficiently small (≈ 10-6), the second term onthe right-hand side of (19) can be significantly weakerthan the first term. (19) also implies that the accumulatedresidual interference from all the ELs may dramaticallyreduce the BL’s performance. As a result, further powerdistribution scheme is definitely required to optimize theoverall system performance as we will discuss later in thefollowing sections.Based on the interference-suppressed signal S’’, hard

decisions are made upon S’’ and the original data onthe BL can be obtained.

IV. Performance evaluation of the proposedsystemA. Error performance analysis of the ELsAs the proposed MCSK modulation is very similar toconventional M-ary orthogonal signaling, the BERexpression given by [33, Sec. 4.4-1] can be used to evalu-ate the performance of the proposed MCSK. However, itis difficult to obtain a closed-form expression and thusextremely high computational complexity for the poten-tial power distribution design is expected if the exactBER expression is used as a constraint. Considering theefficiency of further power distribution scheme, a simpleBER upper bound is derived in this subsection.In order to evaluate the robustness of the proposed

modulation scheme, the peak-to-noise ratio (PNR) of thecorrelation output is first analyzed. For analytical simpli-city, the correlation output in (14) can be rewritten as

R(ϕ) ={A + n if Om = Oϕ

n if Om �= Oϕ(20)

where A =√Pj∑M−1

k=0

∣∣∣Z(j)Om

(k)∣∣∣2 =

√PjM and

n =∑(m+1)M−1

k=mM[X(k) +W ′(k)

](Z(j)Oϕ(k))∗

denote the

ideal peak gain and the associated interference andnoise term, respectively. n can be considered a Gaussianrandom variable with the distribution,


Page 7 of 18

n ∼ CN (0, σ 2

w

), (21)

where σ 2w = Mσ 2

d +∑(m+1)M−1

mMσ 2n

|H(k)|2. For analytical

simplicity, we consider the signal propa-gation environ-ments where one dominant path exists in the multipathchannel and the maximum channel delay spread is shortwhen compared with the OFDM symbol duration. In thiscase, the channel frequency response can have relativelylow frequency selectivity and σ 2

w can be approximated to

σ 2w ≈ Mσ 2

d +Mσ 2n

∣∣H(0)∣∣2 = Mσ 2

d (1 + γ ), (22)

where γ � σ 2n

∣∣H(0)∣∣2σ 2

dis defined as the inverse SNR.

Therefore the PNR of the correlation output can berepresented by

PNRj �A2

σ 2w

=PjM2

Mσ 2d (1 + γ )

. (23)

It is worth mentioning that in the case of (9), whereeach segment on the EL is repeated by R times forrobustness enhancement, the corresponding segmentscan be coherently combined prior to correlation detec-tion. The correlation output in (14) can be rewritten as

R(ϕ) =√Pj

M−1∑k=0

∣∣∣z(j)Om(k)∣∣∣2 + M−1∑

k=0

[X(k) +W ′(k)

](Z(j)Oϕ(k))∗

if Om = Oϕ (24)

where X(k) is the averaged OFDM symbol with var-iance σ 2

d /R and W ′(k) is the averaged noise with var-iance σ 2

w/R. The corresponding PNR can then bereformulated by,

PNRj =PjM2

Mσ 2d (1 + γ )

.R. (25)

As we can see from (20), the correct detection of thecyclic phase shift which matches the maximum correla-tion output should meet the criterion A > n + n for theM − 1 comparisons.Now let us consider a new variable y = 2n and its

probability density can be derived as follows

pY(y) =

∞∫−∞

pN(n)pN(y − n)dn

=

∞∫−∞

1√2πσ 2

w

e− n2

2σ 2w

1√2πσ 2

w

e− (y−n)2

2σ 2w dn

=1√2πσ 2

w

e− y2

2σ 2w

∞∫−∞

1√2πσ 2

w

e−2(n−y/2)2−y2/2

2σ 2w

=1

2√

πσwe− y2

4σ 2w .

(26)

The correct detection should meet the criteria thaty < A for all the M − 1 comparisons of the correlationoutput and therefore, the false detection probability forone comparison is given by

Pc =

∞∫A

1

2√

πσwe− y2

4σ 2w dy

= Q( A√

2σw

)= Q

(√PNR2

).

(27)

The overall error probability of peak detection is thenupper bounded by its union bound,

Pf = (M − 1) Pc. (28)

Assume that all the data bits in one data symbol di areequally likely. Since one symbol is composed of k bits,the BER of the jthe EL is therefore upper bounded by

Pb,j =2k−1

2k−1Pf

= 2k−1Q(√

Pj · NPNRj

2

)

≤ M − 12

exp(

−Pj · NPNRj

4

),

(29)

where we define the normalized PNR of the jth EL asNPNRj = PNRj/

√Pj.

B. Capacity loss analysis of BLThe average channel capacity of the BL is derived in thissubsection to study the impact of EL’s transmission onthe overall system performance. The effective receivedsignal after interference cancelation can be representedas

Y = X · H + I +W. (30)

Assume that the least square or minimum meansquare error estimators using training sequences areused for channel estimation in the proposed system andif the length of the training sequence is Nt, when L/Nt issufficiently small, the residual interference term I in isessentially uncorrelated with W. Therefore, I + W in(30) can be considered as a Gaussian vector with zeromean and covariance matrix of(∑K

i=1Piσ 2

�H + 2Pe,iPiσ 2H + σ 2

n

)IN, where IN is an iden-

tity matrix of order N. Based on the analysis of the aver-age channel capacity for flat fading channels given in[34], we derive the average channel capacity of the BLCBL by averaging the capacity of each subcarrier over allthe subcarriers,


Page 8 of 18

CBL =1N

N−1∑k=0

E

[log

(1 +

PB · ∣∣H′k

∣∣2PBσ 2

�H +∑K

i=1

(Piσ 2

�H + 2Pe,iPiσ 2H

)+ σ 2

n

)], (31)

where PB denotes the transmit power of the BL. Foranalytical simplicity, by introducing the followingGaussian random variable with zero mean and unit

variance, g � H′k/√Var(H′

k

), (31) can be reformulated

as

CBL =1N

N−1∑k=0

E

⎡⎢⎢⎢⎣log

⎛⎜⎜⎜⎝1 +

PB · Var (H′k

) ∣∣g∣∣2Ptotalσ 2

�H +K∑i=1

2Pe,iPiσ 2H + σ 2

n

⎞⎟⎟⎟⎠⎤⎥⎥⎥⎦,(32)

where Ptotal = PB + PE,total represents the total trans-mit power of the overall system and PE,total =

∑Ki=1 Pi

denotes the total transmit power of the ELs. Further-more, the assumption that the normalized channel esti-mation error σ 2

�H/σ2H is sufficiently small holds when

accurate channel estimation techniques are adopted andtherefore, Var

(H′

k

)can be approximated to σ 2

H. Then thecapacity can be further approximated as

CBL ≈ 1N

N−1∑k=0

E

⎡⎢⎢⎢⎣log

⎛⎜⎜⎜⎝1 +

PB · σ 2H

∣∣g∣∣2Ptotalσ 2

�H +K∑i=1

2Pe,iPiσ 2H + σ 2

n

⎞⎟⎟⎟⎠⎤⎥⎥⎥⎦

= log

(1 +

PB · σ 2H

Ptotalσ 2�H +

∑Ki=1 2Pe,iPiσ

2H + σ 2

n

).

(33)

In the meantime, the upper bound of the BL’s capacityin the absence of ELs’ transmission can be written as

C̄BL =1N

N−1∑k=0

log(1 +

PB · σ 2H

PBσ 2�H + σ 2

n

). (34)

By introducing the maximum allowed capacity lossΔC, the ELs’ transmission is enabled only if the follow-ing constraint is satisfied

C̄BL − CBL ≤ �C. (35)

The above constraint is referred to as the BL’s capa-city loss constraint and can be reformulated as

CBL ≥ C, (36)

where C � C̄BL − �C. The above constraint is essen-tial for the ML-OFDM system design as it reflects theimpact of ELs’ transmission on the BL. If the capacityloss is sufficiently large that the BL cannot tolerate, noELs’ transmission is allowed. Therefore, the constraintwill further be used in the power distribution scheme aswe will discuss in the following section.

V. Power distribution for ELs’ transmissionPower distribution scheme for the ELs is discussed inthis section. The objective is to optimize the overall sys-tem performance by balancing the tradeoff between theproposed ELs and the BL. Therefore, we propose tominimize the overall BER of the ELs with the constrainton the BL’s capacity loss. Furthermore, by taking thedifferent reliability/coverage of different ELs intoaccount, we introduce the proportional reliability con-straints into our system. The benefit of these constraintsis that we can flexibly control the reliability of differentELs for different purposes and therefore, ensuring thateach signaling is able to achieve its target service qualitygiven sufficient available transmit power and tolerableBL’s capacity loss.The power distribution problem can be expressed

mathematically as,

minPi

BER =

∑Ki=1 kiPb,i∑Ki=1 ki

, (37)

subject to,

∑K

i=1Pi ≤ PE,total (38)

CBL ≥ C (39)

Pb,1 : Pb,2 : · · · : Pb,K = γ1 : γ2 : · · · : γK , (40)

where Pb,i denotes the BER upper bound of the ith ELas derived in (29). ki denotes the total effective numberof transmitted bits of the ith EL. PE,total is the totaltransmit power budget allocated to the ELs. {γi}Ki=1 is theset of the predefined values which are used for the pro-portional reliability constraints. Note that the nonlinearinequality constraint -CBL ≤ -C makes the optimizationproblem in (37) nonconvex. Iterative methods, such asNewton-Raphson or quasi-Newton methods can be usedto obtain the solutions; however, with a large amount ofcomputational complexity. Fortunately, under certainapproximations, the optimization problem can berelaxed to convex problem and therefore, the optimal ornear-optimal solutions of problem can be found withlow complexity. Due to the MCSK used on ELs, theoperating SNR range for the ELs is much lower thanthat of the BL. Therefore, we analyze the low SNR casewhere certain approximations can be made.Under relatively low SNR range, the SER of the ELs

Pe,i, i = 1,2,...,K are already very low and significantlysmaller than the variance of channel estimation error

σ 2�h. The term

K∑i=1

2Pe,iPiσ 2H in (33) can be neglected and


Page 9 of 18

the capacity of BL given by (33) can further be simpli-fied to

CBL ≈ log

(1 +

PB · σ 2H

PBσ 2�H +

∑Ki=1 Piσ

2�H + σ 2

n

). (41)

It is straightforward that the approximated capacity isnow a concave function w.r.t. Pi, which makes the opti-mization problem convex. Its global optimal solutioncan then be obtained by Karush-Kuhn-Tucker (KKT)conditions [35] as follows,

∂BER∂Pi

+ λ + μ∂CBL

∂Pi+

K−1∑i=1

i

∂(Pb,1 − γ1

γiPb,i)

∂Pi= 0 (42)

λ

(K∑i=1

Pi − PE,total

)= 0 (43)

μ

(log

(1 +

PB · σ 2H

PBσ 2�H +

∑Ki=1 Piσ

2�H

)+ C

)= 0 (44)

i

(exp

(−P1 · NPNR1

4

)− γ1

γiexp

(−Pi · NPNRi

4

))= 0, i = 2, ...,K. (45)

where l, μ, and ϖi are the Lagrange multipliers. l ≥ 0,μ ≥ 0, and ϖi ≥ 0 ∀i Î {1,...,K}.Note that the capacity given by (33) is a monoto-

nously increasing function w.r.t. Pi and therefore, theconstraint in (39) can be reformulated by

K∑i=1

Pi ≤ PBσ 2H(

10C − 1)σ 2

�H

− PB − σ 2n

σ 2�H

� χ , (46)

where the right-hand side of the above inequality isdenoted by c. The condition in (44) can then be rewrit-ten by

μ

(K∑i=1

Pi − χ

)= 0 (47)

From (43) and (47), we note that l and μ are notallowed to be synchronously nonzero which means

λ �= 0 ⇒K∑i=1

Pi = χ (48)

μ �= 0 ⇒K∑i=1

Pi = PE,total, (49)

when c ≠ Ptotal. Therefore, the optimization problemcan be discussed in the following circumstances:

A. When c < PE,totalWith this condition, we can obtain

∑Ki=1 Pi < PE,total

since∑K

i=1 Pi ≤ χ. According to (43), we must attain l =0. Therefore, (42) can furthered be expanded and solvedby

kiK∑i=1

ki

·(

−NPNRi

4

)Pb,i + μ + i

(−γ1

γi

)·(

−NPNRi

4

)Pb,i = 0

⇒ Pi =[− 4NPNRi

log{

8μ

ηi (M − 1)NPNRi

}, 0]+,

(50)

where [x, y]+ ≜ max {x, y} andηi � ki

K∑k=1

ki

− γ1γi

i. Note

that when μ = l = 0, then it is easy to obtain Pi ® ∞,which is impossible for real implementation. Therefore,this circumstance is not allowed to occur. μ and l mustnot be synchronously zero.

B. When c ≥ PE,totalSimilarly, we can obtain that

∑Ki=1 Pi < χ since∑K

i=1 Pi ≤ Ptotal. Thus, μ = 0 in this circumstance. Thesolution is thus given by with μ replaced by l

Pi =[− 4NPNRi

log{

8λ

ηi (M − 1)NPNRi

}, 0]+. (51)

Also μ and l are not allowed to be synchronouslyzero.From the optimal power distribution solution for the

ELs, it implies that the power level for different ELsdepends on the parameters l, μ, and ϖi, i = 2,..., K.First, l is the dual variable associated with the totaltransmit power budget. It is straightforward that a largertransmit power budget will result in a smaller l andthus a high power level, and vice versa. Second, μ is thedual variable associated with the tolerable capacity lossof the BL. If the BL can accommodate a larger residualinterference introduced by the transmission of ELs, μwould be smaller, and therefore a higher power level,and vice versa. For instance, in an extreme case wherethe BL cannot accommodate any interference or inother words, the capacity loss constraint for the BL iszero, then μ would be approaching infinity and theresultant zero power level indicates that no ELs’ trans-mission is allowed in this condition. Similarly, the analy-sis can be also applied to ϖi, i = 2,..., K which areassociated with the proportional reliability constraintsfor the ELs.

VI. Simulation results and discussionsComputer simulations have been carried out to evaluatethe performance of the proposed ML-OFDM system.The OFDM system with 1024 subcarriers and CP of


Page 10 of 18

length 1/8 of the symbol duration is considered in thesimulation. The modulation scheme for the BL is chosenas 4-QAM. In addition, two ELs are considered in thedemonstration system which are designated to the mul-ticell cooperation signaling and the MBMS, respectively.According to Section 2.2, two nearly orthogonal Zadoff-Chu sequences with length M = 64 are used as the sig-nature sequences for the two ELs. To show the flexibilityof the ML-OFDM system, we assume that the Globalvector quantization approach [32] is adopted to reducethe overhead of the shared CSI/user data between thecollaborative BSs, and therefore, the requirement of datarate for the multicell cooperation signaling is very low.For power-saving purpose, each data symbol is repeatedby 8 times for the first EL and then the effective numberof transmitted bits becomes k1 = 12bits. The second ELis intended to provide MBMS at a relatively high data

rate, e.g., mobile TV. The signal structures of the twoELs can then be formulated according to

D(1) =

⎡⎢⎣d(1)

0 , ...,d(1)0︸︷︷︸

8

,d(1)1 , ...,d(1)

1︸︷︷︸8

⎤⎥⎦ ,

D(2) =[d(2)0 , ...,d(2)

1 , ...,d(2)14 ,d

(2)15

].

(52)

A. The impact of BL’s capacity loss constraintWe first examine the impact of the BL’s capacity lossconstraint on the ELs’ transmission. Figure 4 shows theanalytical results of the maximum allowed total transmitpower for the ELs with different BL’s capacity loss con-straints. It can be observed that as the capacity loss con-straint becomes looser, higher transmit power can be

0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.20

5%

10%

15%

20%

25%

30%

35%

40%

45%

Variance of channel estimation error σΔH2

Cap

acity

loss

of B

L

SNR 0dBSNR 5dBSNR 10dBSNR 20dB

Figure 4 Impact of the varying BL’s capacity loss constraints on the maximum ELs’ transmit power with fixed channel estimationerror σ 2

�H = 0.10.


Page 11 of 18

allocated to the ELs, which will significantly improve thereliability/coverage of each EL; however, the perfor-mance of cellular communications will be correspond-ingly degraded. It is also apparent that an increase inSNR will result in a decrease in the attained total trans-mit power. This is because as the variance of theAWGN reduces, the capacity loss is dominated by the

term∑K

i=1Pi · σ 2

�H. If there exists large channel estima-

tion error, then a small variation in PE,total will causedramatic reduction in BL’s capacity and therefore noELs’ transmission is beneficial in this case. Therefore,we subsequently assess the consequences of differentaccuracy of channel estimation on the BL’s capacity.From Figure 5, it is observed that the capacity lossincreases as the channel estimation becomes less accu-rate. In particular, at the SNR level of 0 dB, when thevariance of channel estimation error is larger than 0.1,the capacity loss can be larger than 35%.

B. Power distribution using the proposed algorithm forthe ELsTo show the near-optimality of the proposed power dis-tribution in Section 5, the solution is compared with anoptimal power distribution scheme under different accu-racy of channel estimation in Figure 6. We assume thatthe EL’s total available transmit power budget is suffi-ciently large such that the optimization problem is con-strained by the BL’s capacity loss and the ELs’proportional reliability. We further set the BL’s capacityloss to 10% and proportional reliability constraint to g1 :g2 = 1 : 2. The optimal power distribution solves theproblem by using the exact BER expressions of the ELs[33] instead of the derived BER upper bounds. Theexact BER expression can be given by

Pb,i =2k−1

2k − 1· 1√

2π

∞∫−∞

[1 − (1 − Q(x))M−1

]exp

(−(x − √

PNRi)2

2

)dx. (53)

0 2 4 6 8 10 12 14 16 18 200

30

60

90

120

150

SNR (dB)

Max

imum

ELs

' tra

nsm

it po

wer

BL's capacity loss 5%BL's capacity loss 10%BL's capacity loss 20%BL's capacity loss 30%BL's capacity loss 50%

Figure 5 Impact of the accuracy of channel estimation on the BL’s capacity loss with fixed ELs’ total transmission power PE,total = 1.


Page 12 of 18

The above expressions can only be calculated numeri-cally and therefore, an exhaustive searching must be car-ried to to find the optimal solution P1 and P2 for thetwo ELs with extremely high computational complexity.Figure 6 shows that the gap between the proposed andthe optimal power distribution is almost invisible andtherefore, it confirms the near-optimality of the pro-posed power distribution.The BERs of the ELs obtained by the proposed power

distribution are plotted in Figure 7. It can be seen thatthe BERs of the two ELs are well differentiated accord-ing to the proportional constraints. For comparison, wealso show the BERs derived from equal power distribu-tion and the conventional optimal power distributionwithout the proportional reliability constraints [36]. It

can be observed that for equal power distribution, dueto the different data rate (hence different PNRs), thefirst EL achieves too much gain in BER over the secondEL which leads to unfair resource allocation. Meanwhile,for the conventional waterfilling scheme in [36], it tendsto allocate power such that the performance of the twoELs are similar. It is worth mentioning that the BERratio of the two ELs may not be strictly equivalent tothe predefined proportional constraints due to the useof the BER upper bounds; however, large computationalburden is reduced which is more meaningful to realisticcellular network scenario.To evaluate more intuitively how good the proposed

power distribution scheme satisfies the proportionalreliability constraints, a new metric is defined as follows.

0.5 1 1.5 2 2.50

2

4

6

8

10

Variance of channel estimation error σ2

Allo

cate

d po

wer

Allocated power of EL #1 - ProposedAllocated power of EL #1 - OptimalAllocated power of EL #2 - ProposedAllocated power of EL #2 - Optimal

ΔHFigure 6 The allocated power to different ELs with the proposed power distribution scheme and the optimal power distributionscheme. The BL’s capacity loss is 10% and the ELs’ proportional reliability constraint is g1 : g2 = 1 : 2. The total available transmit power budget

for the ELs is assumed to be sufficiently large. The variance of channel estimation error is σ 2�H = 0.1. The optimal power distribution uses the

exact BER expressions given by [33] instead of the derived BER upper bounds for the proportional reliability constraints.


Page 13 of 18

Let Pkb,i be the BER of the ith EL for the kth Monte

Carlo run and P̃kb,i = Pk

b,i/∑K

i=1Pkb,i be the normalized

BER of the ith EL. Also, the normalized proportionalreliability constraint is defined as γ̃i = γi/

∑Ki=1 γi. Then

the variance of the reliability proportion for the kthMonte Carlo run is defined as

Vk =K∑i=1

∣∣∣P̃kb,i − γ̃i

∣∣∣2. (54)

The average deviation over total I Monte Carlo runs,denoted by D =

∑Ik=1

√Vk/I, is reported in Table 2.Note that the ideal D is supposed to be close to zero ifthe allocated power strictly satisfies the constraints. Itcan be observed that the deviations of the reliability pro-portion obtained by the proposed power distribution

scheme is orders of magnitude smaller than thoseobtained by equal power distribution and conventionalwaterfilling algorithm.

C. The impact of ELs’ transmission on BLAfter the power distribution of the ELs, the SER of theBL which indicates the service quality of cellular com-munications is examined in the presence of ELs’ trans-mission with different channel estimation accuracy inFigure 8. The curve labeled “ideal coherent detection”refers to the SER obtained by perfect channel estimationand interference cancelation. When highly accuratechannel estimation scheme is used, the SER degradationof the BL is almost indistinguishable which indicatesthat the proposed ELs for multicell cooperation signalingand other purposes have virtually no impact on the cel-lular users.

-24 -22 -20 -18 -16 -14 -1210-6

10-5

10-4

10-3

10-2

10-1

100

SNR (dB)

BE

R

EL #1 - con waterfilling in [36]EL #2 - con waterfilling in [36]EL #1 - equal power distributionEL #2 - equal power distributionEL #1 - proposed power distributionEL #2 - proposed power distribution

Figure 7 BERs of the ELs by using the proposed power distribution, equal power distribution and the conventional power distributionwithout the proportional reliability constraints. The parameters of the constraints are the same as Figure 6.


Page 14 of 18

D. An option of channel estimation for accuracyimprovementIt can be seen from the previous simulation results thatthe accuracy of channel estimation has large impact onthe overall system performance. In slot transmission-based cellular networks, a preamble signal is periodicallyinserted in the transmitted data stream for initial syn-chronization and channel estimation purposes. Usuallythe overhead of the preamble signals should be keptsmall in order to the improve the transmission efficiencyof the network and therefore, the accuracy of channelestimation can be limited by the length of the preamble

signal. We propose an iterative decision-directed chan-nel estimation where the accuracy of initial estimation isimproved with the aid of the detected data. The processis briefly described as follows: First, the initial channelestimate is obtained by the preamble signal, which isthen applied to the coherent detection of the subsequentOFDM symbols. The time-domain transmitted signal isthen regenerated by IDFT and combined with the origi-nal preamble signal to formulate an extended preamblesuch that the length and total power of this extendedpreamble is substantially improved. This new “preamble”is utilized to update the channel estimation with

Table 2 Deviations of the proportional reliability constraints under different SNRs

SNR(dB) -24 -22 -20 -18 -16 -14 -12

g1 : g2 0.5 0.5 0.5 0.5 0.5 0.5 0.5

D, proposed power distribution 0.0012 0.0052 0.0092 0.0217 0.0170 0.0021 0.0476

D, equal power distribution 0.0065 0.0651 0.1743 0.2162 0.2222 - -

D, water filling algorithm in [36] 0.0535 0.0527 0.0515 0.0603 0.0628 0.0127 0.1591

0 5 10 15 20

10-2

10-1

SNR (dB)

SE

R

Var of CE error σΔH2 =0.1





ideal coherent detection

Figure 8 Impact of ELs’ transmission on BL’s performance with different accuracy of channel estimation.


Page 15 of 18

improved accuracy. The above process is iterated tosimultaneously provide more accurate channel and dataestimation.Figure 9 presents the mean square error of the itera-

tive decision-directed channel estimation under a 10-tapmultipath channel with uniform power delay profile andthe total channel energy σ 2

H is normalized to 1. The pre-amble is a pseudo random sequence with 40 samples. InFigure 9, the curve labeled “0 iteration” represents theconventional channel estimation by utilizing only thepreamble signal. The lower bound represents the casewhere the transmitted signal is perfectly regeneratedand subsequently used as part of the extended preamble.It can be observed that the iterative decision-directedchannel estimation substantially outperforms the con-ventional preamble-based channel estimation. With an

increase in iterations, the mean square error of channelestimation reduces and achieves the lower bound atapproximately 20 dB.

VII. ConclusionA new ML-OFDM supporting multicell cooperative net-works is proposed in this article. Besides the basic func-tion of the BL as a multiple access technique for thecellular users, multiple functionalities for multicell coop-eration purposes are derived from overlay ELs. Byencoding the required information into these layers,concurrent transmission of cooperation-related para-meters can be achieved with the cellular users’ data, andtherefore, complicated procedure to establish backhaulnetwork or additional signaling can be substantially sim-plified. Cross BS synchronization can also be obtained by

0 5 10 15 20 2510-5

10-4

10-3

10-2

10-1

100

SNR (dB)

Mea

n sq

uare

err

or

0 iteration1 iteration2 iterations3 iterationslower bound

Figure 9 Mean square error of channel estimation using the proposed iterative decision-directed scheme. A slot transmission mode isassumed where a preamble signal is periodically multiplexed with data-carrying OFDM symbols for initial synchronization and channelestimation purposes. A pseudo random sequence with 40 samples is used as the preamble signal. A 10-tap multipath channel with normalizeduniform average power delay profile is used.


Page 16 of 18

the proposed EL, which significantly reduce the com-plexity in the design of signal transmission strategies formulticell cooperation. The corresponding transceiver isdesigned for the proposed ML-OFDM system based onthe proposed multi-layer interference cancelation. Prac-tical power distribution scheme is also proposed to opti-mize the overall system performance. The performanceof the ML-OFDM is theoretically analyzed and verifiedthrough computer simulations. With the ML-OFDMplatform, multicell cooperation as well as various wire-less demands will become more efficient and flexibleand can be easily achieved.

EndnotesaFor instance, the broadcast capabilities are achieved byallocating dedicated broadcast time slots in WiMAX[37,38].

AbbreviationsAWGN: additive white Gaussian noise; BER: bit error rate; BL: base layer; BS:base station; CP: cyclic prefix; CSI: channel state information; CoMP:coordinated multi-point; EL: enhanced layer; EUTRA: evolved universalterrestrial radio access; FDM: frequency-division multiplexing; GPS: globalpositioning system; IDFT: inverse discrete Fourier transform; ISI: intersymbolinterference; KKT: Karush-Kuhn-Tucker; MAC: media access control; MBMS:multimedia broadcast multicast service; MCSK: modified code shift keying;MIMO: multiple-input-multiple-output; ML-OFDM: multi-layered orthogonalfrequency division multiplexing; NPNR: normalized peak-to-noise ratio;OFDMA: orthogonal frequency division multiple access; PHY: physical; PNR:peak-to-noise ratio; PRACH: random access preamble; PSS: primarysynchronization signal; QAM: quadrature amplitude modulation; SCP: singlecell processing; SER: symbol error rate; SNR: signal-to-noise ratio; TDM: time-division multiplexing.

AcknowledgementsThis research was supported in part by Natural Sciences and EngineeringResearch Council of Canada (NSERC) Discovery Grant R4122A02 and the KCC(Korea Communications Commission), Korea, under the support programsupervised by the KCA (Korea Communications Agency) (09912-02001).

Author details1Bell Centre for Information Engineering, Department of Electrical andComputer Engineering, The University of Western Ontario, London, Ontario,N6A 5B9, Canada 2Broadcasting & Telecommunications ConvergenceResearch Laboratory, Electronics and Telecommunications Research Institute,Daejeon, Republic of Korea

Competing interestsThe authors declare that they have no competing interests.

Received: 6 July 2011 Accepted: 12 March 2012Published: 12 March 2012

References1. U Horn, R Keller, N Niebert, Iterative mobile streaming services-convergence

of broadcast and mobile communication. EBU Techn Rev 14–19 (1999)2. R Keller, T Lohman, R Tonjes, J Thielecke, Covergence of cellular and

broadcast networks from a multi-radio perspective. EEE Perspec Commun.8(2), 51–56 (2001)

3. O Benali, KE -Khazen, D Garrec, M Guiraudou, G Martinez, A framework foran evolutionary path toward 4G by means of cooperation of networks. IEEECommun Mag. 42(5), 82–89 (2004)

4. F Allamandri, S Campion, A Centonza, A Chernilov, JP Cosmas, A Duffy, DGarrec, M Guiraudou, K Krishnapillai, T Levesque, B Mazieres, R Mies, T

Owens, Re M, E Tsekleves, L Zheng, Service platform for convergedinteractive broadband broadcast and cellular wireless. IEEE Trans Broadcast.53(1), 200–211 (2007)

5. J Andrews, Interference cancellation for cellular systems: A contemporaryoverview. IEEE Wirel Commun. 12(2), 19–29 (2005). doi:10.1109/MWC.2005.1421925

6. D Gesbert, S Hanly, H Huang, S Shamai, O Simeone, Yu W, Multi-cell MIMOcooperative netwokrs: A new look at interference. IEEE J Sel Area Commun.28(9), 1380–1408 (2010)

7. S Shamai, B Zaidel, Enhancing the cellular downlink capacity via co-processing at the transmitter end, in Proc IEEE 53rd Veh Technol Conf(VTC’01 Spring), vol. 3. (Rhodes, Greece, 2001), pp. 1745–1749

8. M Karakayali, G Foschini, R Vlenzuela, Network coordination for spectrallyefficient communications in cellular systems. IEEE Wirel Commun. 13(4),56–61 (2006). doi:10.1109/MWC.2006.1678166

9. H Zhang, H Dai, Cochannel interference mitigation and cooperativeprocessing in downlink multicell multiuser MIMO networks. EURASIP J WirelCommun Netw. 2004(2), 222–235 (2004)

10. B Clerckx, A Lozano, S Sesia, CV Rensburg, CB Papadias, 3 GPP LTE and LTEAdvanced. EURASIP J Wirel Commun Netw. 2009, 1–3 (2009). Article ID472124

11. H Weingarten, Y Steinberg, S Shamai, The capacity region of the Gaussianmultiple-input-multiple-output broadcast channel. IEEE Trans Inf Theory.52(9), 3936–3964 (2006)

12. O Somekh, B Zaidel, S Shamai, Sum rate characterization of joint multiplecell-site processing. IEEE Trans Inf Theory. 53(12), 4473–4497 (2007)

13. O Somekh, S Shamai, Shannon-theoretic approach to a Gaussian cellularmultiple access channel with fading. IEEE Trans Inf Theory. 46(4), 1401–1425(2000). doi:10.1109/18.850679

14. N Levy, O Zeitouni, S Shamai, Information theoretic aspects of users’ activityin a Wyner-like cellular model. IEEE Trans Inf Theory. 56(5), 2241–2248(2010)

15. Y Liang, A Goldsmith, Symmetric rate capacity of cellular systems withcooperative base stations, in Proc IEEE Globe Telecommu Conf 2006(GLOBECOM ‘06), (San Francisco, USA, 2006), pp. 1–5

16. M Bacha, J Evans, S Hanly, On the capacity of cellular networks with mimolinks, in Proc IEEE Int Conf Commun 2006 (ICC’06), vol. 3. (Istanbul, Turkey,2006), pp. 1337–1342

17. G Caire, S Shamai, On the achievable throughput of a multi-antennaGaussian broadcast channel. IEEE Trans Inf Theory. 49(7), 1691–1706 (2003).doi:10.1109/TIT.2003.813523

18. O Simeone, O Somekh, HV Poor, S Shamai, Downlink multicell processingwith limited backhaul capacity. EURASIP J Adv Signal Process. 2009, 1–10(2009). Article ID 840814

19. O Simeone, O Somekh, Y Bar-Ness, U Spagnolini, Throughput of low-powercellular systems with collaborative base stations and relaying. IEEE Trans InfTheory. 54(1), 459–467 (2008)

20. S Jing, DNC Tse, JB Soriaga, J Hou, JE Smee, R Padovani, Multicell downlinkcapacity with coordinated processing. EURASIP J Wireless Commun Netw.2008, 1–19 (2008). Article ID 586878

21. O Simeone, O Somekh, G Kramer, HV Poor, S Shamai, Throughput ofcellular systems with conferencing mobiles and cooperative base stations.EURASIP J Wireless Commun Netw. 2008, 1–14 (2008). Article ID 652325

22. H Skjevling, D Gesbert, A Hjorungners, Low-complexity distributedmultibase transmission and scheduling. EURASIP J Adv Signal Process. 2008,1–9 (2008). Article ID 741593

23. W Choi, J Andrews, The capacity gain from base station cooperativescheduling in a MIMO DPC cellular system, in Proc IEEE Int Symp Inf Theory,(Seattle, USA, 2006), pp. 1224–1228

24. W Yu, T Kwon, C Shin, Multicell coordination via joint scheduling,beamforming and power spectrum adaptation, in Proc IEEE INFOCOM 2011,(Shanghai, P.R.China, 2011), pp. 2570–2578

25. SG Kiani, D Gesbert, Optimal and distributed scheduling for multicellcapacity maximization. IEEE Trans Wirel Commun. 7(1), 288–297 (2008)

26. T Hwang, C Yang, G Wu, S Li, GY Li, OFDM and its wireless applications: asurvey. IEEE Trans Veh Technol. 58(4), 1673–1694 (2009)

27. YK Khit, TT Jhung, Concatenated sequences for inter-base stationsynchronization in WCDMA TDD mode, in Proc IEEE 8th Int Conf CommunSys 2002 (ICCS’02), vol. 1. (Singapore, 2002), pp. 446–450


Page 17 of 18

28. GM Dillard, M Reuter, J Zeidler, B Zeidler, Cyclic code shift keying: a lowprobability of intercept communication technique. IEEE Trans Aero ElectronSyst. 39(3), 786–798 (2003). doi:10.1109/TAES.2003.1238736

29. S Erkucuk, DI Kim, KS Kwak, Code shift keying impulse modulation for UWBcommunications. IEEE Trans Wirel Commun. 7(9), 3285–3291 (2008)

30. YR Tsai, X-S Li, Kasami code-shift-keying modulation for ultra widebandcommunication systems. IEEE Trans Commun. 55(6), 1242–1252 (2007)

31. R Bhagavatula, R Heath, Adaptive bit partitioning for multicell intercellinterference nulling with delayed limited feedback. IEEE Trans SignalProcess. 59(8), 3824–3836 (2011)

32. Y Liang, R Schober, W Gerstacker, Time-domain transmit beamforming forMIMO-OFDM system with finite rate feedback. IEEE Trans Commun. 57(9),2828–2838 (2009)

33. JG Proakis, Digital Communications, (McGraw-Hill, New York, 1995)34. M Medard, The effect upon channel capacity in wireless communications of

perfect and imperfect knowledge of the channel. IEEE Trans Inf Theory.46(3), 933–946 (2000). doi:10.1109/18.841172

35. S Boyd, L Vandenberghe, Convex Optimization, (Cambridge University Press,Cambridge, UK, 2004)

36. CS Park, KB Lee, Transmit power allocation for BER performanceimprovement in multicarrier systems. IEEE Trans Commun. 52(10),1658–1663 (2004). doi:10.1109/TCOMM.2004.836491

37. T Jiang, W Xiang, HH Chen, Q Ni, Multicast broadcast services support inOFDMA-based WiMAX systems. IEEE Commun Mag. 45(8), 78–86 (2007)

38. K Etemad, L Wang, Multicast and broadcast multimedia services in mobileWiMax networks. IEEE Commun Mag. 47(10), 84–91 (2009)

doi:10.1186/1687-1499-2012-103Cite this article as: Yang et al.: A multi-layered OFDM system withparallel transmission for multicell cooperative cellular networks. EURASIPJournal on Wireless Communications and Networking 2012 2012:103.

Submit your manuscript to a journal and benefi t from:

7 Convenient online submission

7 Rigorous peer review

7 Immediate publication on acceptance

7 Open access: articles freely available online

7 High visibility within the fi eld

7 Retaining the copyright to your article

Submit your next manuscript at 7 springeropen.com


Page 18 of 18

http://www.springeropen.com/

http://www.springeropen.com/

Documents

RESEARCH Open Access A multi-layered OFDM system with ......RESEARCH Open Access A multi-layered OFDM system with parallel transmission for multicell cooperative cellular networks